| Message ID | 20231024064416.897956-4-hch@lst.de (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [1/3] filemap: add a per-mapping stable writes flag | expand |
On Tue, Oct 24, 2023 at 08:44:16AM +0200, Christoph Hellwig wrote: > Update the per-folio stable writes flag dependening on which device an > inode resides on. > > Signed-off-by: Christoph Hellwig <hch@lst.de> > --- > fs/xfs/xfs_inode.h | 8 ++++++++ > fs/xfs/xfs_ioctl.c | 9 +++++++++ > fs/xfs/xfs_iops.c | 7 +++++++ > 3 files changed, 24 insertions(+) > > diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h > index 0c5bdb91152e1c..682959c8f78cb0 100644 > --- a/fs/xfs/xfs_inode.h > +++ b/fs/xfs/xfs_inode.h > @@ -561,6 +561,14 @@ extern void xfs_setup_inode(struct xfs_inode *ip); > extern void xfs_setup_iops(struct xfs_inode *ip); > extern void xfs_diflags_to_iflags(struct xfs_inode *ip, bool init); > > +static inline void xfs_update_stable_writes(struct xfs_inode *ip) > +{ > + if (bdev_stable_writes(xfs_inode_buftarg(ip)->bt_bdev)) > + mapping_set_stable_writes(VFS_I(ip)->i_mapping); > + else > + mapping_clear_stable_writes(VFS_I(ip)->i_mapping); > +} > + > /* > * When setting up a newly allocated inode, we need to call > * xfs_finish_inode_setup() once the inode is fully instantiated at > diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c > index 55bb01173cde8c..67bf613b3c86bc 100644 > --- a/fs/xfs/xfs_ioctl.c > +++ b/fs/xfs/xfs_ioctl.c > @@ -1147,6 +1147,15 @@ xfs_ioctl_setattr_xflags( > ip->i_diflags2 = i_flags2; > > xfs_diflags_to_iflags(ip, false); > + > + /* > + * Make the stable writes flag match that of the device the inode > + * resides on when flipping the RT flag. > + */ > + if (S_ISREG(VFS_I(ip)->i_mode) && > + XFS_IS_REALTIME_INODE(ip) != (fa->fsx_xflags & FS_XFLAG_REALTIME)) > + xfs_update_stable_writes(ip); Hmm. Won't the masking operation here result in the if test comparing 0 or FS_XFLAG_REALTIME to 0 or 1? Oh. FS_XFLAG_REALTIME == 1, so that's not an issue in this one case. That's a bit subtle though, I'd have preferred XFS_IS_REALTIME_INODE(ip) != !!(fa->fsx_xflags & FS_XFLAG_REALTIME)) to make it more obvious that the if test isn't comparing apples to oranges. > + > xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG); > xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); > XFS_STATS_INC(mp, xs_ig_attrchg); > diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c > index 2b3b05c28e9e48..b8ec045708c318 100644 > --- a/fs/xfs/xfs_iops.c > +++ b/fs/xfs/xfs_iops.c > @@ -1298,6 +1298,13 @@ xfs_setup_inode( > gfp_mask = mapping_gfp_mask(inode->i_mapping); > mapping_set_gfp_mask(inode->i_mapping, (gfp_mask & ~(__GFP_FS))); > > + /* > + * For real-time inodes update the stable write flags to that of the RT > + * device instead of the data device. > + */ > + if (S_ISREG(inode->i_mode) && XFS_IS_REALTIME_INODE(ip)) > + xfs_update_stable_writes(ip); I wonder if xfs_update_stable_writes should become an empty function for the CONFIG_XFS_RT=n case, to avoid the atomic flags update? (The extra code is probably not worth the microoptimization.) With the !! nit addressed, Reviewed-by: Darrick J. Wong <djwong@kernel.org> --D > + > /* > * If there is no attribute fork no ACL can exist on this inode, > * and it can't have any file capabilities attached to it either. > -- > 2.39.2 >
On Tue, Oct 24, 2023 at 08:09:04AM -0700, Darrick J. Wong wrote: > > + if (S_ISREG(VFS_I(ip)->i_mode) && > > + XFS_IS_REALTIME_INODE(ip) != (fa->fsx_xflags & FS_XFLAG_REALTIME)) > > + xfs_update_stable_writes(ip); > > Hmm. Won't the masking operation here result in the if test comparing 0 > or FS_XFLAG_REALTIME to 0 or 1? > > Oh. FS_XFLAG_REALTIME == 1, so that's not an issue in this one case. > That's a bit subtle though, I'd have preferred > > XFS_IS_REALTIME_INODE(ip) != !!(fa->fsx_xflags & FS_XFLAG_REALTIME)) > > to make it more obvious that the if test isn't comparing apples to > oranges. != !! might be going a bit far. Would you settle for XFS_IS_REALTIME_INODE(ip) == !(fa->fsx_xflags & FS_XFLAG_REALTIME) ? Although none of these read particularly nicely. Maybe XFS_IS_REALTIME_INODE(ip) != ((fa->fsx_xflags & FS_XFLAG_REALTIME) == 0)) Perhaps we need a bool helper for (fa->fsx_xflags & FS_XFLAG_REALTIME)
On Tue, Oct 24, 2023 at 08:09:04AM -0700, Darrick J. Wong wrote: > > + /* > > + * Make the stable writes flag match that of the device the inode > > + * resides on when flipping the RT flag. > > + */ > > + if (S_ISREG(VFS_I(ip)->i_mode) && > > + XFS_IS_REALTIME_INODE(ip) != (fa->fsx_xflags & FS_XFLAG_REALTIME)) > > + xfs_update_stable_writes(ip); > > Hmm. Won't the masking operation here result in the if test comparing 0 > or FS_XFLAG_REALTIME to 0 or 1? > > Oh. FS_XFLAG_REALTIME == 1, so that's not an issue in this one case. > That's a bit subtle though, I'd have preferred > > XFS_IS_REALTIME_INODE(ip) != !!(fa->fsx_xflags & FS_XFLAG_REALTIME)) > > to make it more obvious that the if test isn't comparing apples to > oranges. This is all copy and pasted from a check a few lines above :) I guess I could clean it up as well or even add a rt_flag_changed local variable instead of duplicating the check. > > + /* > > + * For real-time inodes update the stable write flags to that of the RT > > + * device instead of the data device. > > + */ > > + if (S_ISREG(inode->i_mode) && XFS_IS_REALTIME_INODE(ip)) > > + xfs_update_stable_writes(ip); > > I wonder if xfs_update_stable_writes should become an empty function for > the CONFIG_XFS_RT=n case, to avoid the atomic flags update? > > (The extra code is probably not worth the microoptimization.) The compiler already eliminates the code because XFS_IS_REALTIME_INODE(ip) has a stub for CONFIG_XFS_RT=n that always returns 0.
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h index 0c5bdb91152e1c..682959c8f78cb0 100644 --- a/fs/xfs/xfs_inode.h +++ b/fs/xfs/xfs_inode.h @@ -561,6 +561,14 @@ extern void xfs_setup_inode(struct xfs_inode *ip); extern void xfs_setup_iops(struct xfs_inode *ip); extern void xfs_diflags_to_iflags(struct xfs_inode *ip, bool init); +static inline void xfs_update_stable_writes(struct xfs_inode *ip) +{ + if (bdev_stable_writes(xfs_inode_buftarg(ip)->bt_bdev)) + mapping_set_stable_writes(VFS_I(ip)->i_mapping); + else + mapping_clear_stable_writes(VFS_I(ip)->i_mapping); +} + /* * When setting up a newly allocated inode, we need to call * xfs_finish_inode_setup() once the inode is fully instantiated at diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c index 55bb01173cde8c..67bf613b3c86bc 100644 --- a/fs/xfs/xfs_ioctl.c +++ b/fs/xfs/xfs_ioctl.c @@ -1147,6 +1147,15 @@ xfs_ioctl_setattr_xflags( ip->i_diflags2 = i_flags2; xfs_diflags_to_iflags(ip, false); + + /* + * Make the stable writes flag match that of the device the inode + * resides on when flipping the RT flag. + */ + if (S_ISREG(VFS_I(ip)->i_mode) && + XFS_IS_REALTIME_INODE(ip) != (fa->fsx_xflags & FS_XFLAG_REALTIME)) + xfs_update_stable_writes(ip); + xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG); xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); XFS_STATS_INC(mp, xs_ig_attrchg); diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c index 2b3b05c28e9e48..b8ec045708c318 100644 --- a/fs/xfs/xfs_iops.c +++ b/fs/xfs/xfs_iops.c @@ -1298,6 +1298,13 @@ xfs_setup_inode( gfp_mask = mapping_gfp_mask(inode->i_mapping); mapping_set_gfp_mask(inode->i_mapping, (gfp_mask & ~(__GFP_FS))); + /* + * For real-time inodes update the stable write flags to that of the RT + * device instead of the data device. + */ + if (S_ISREG(inode->i_mode) && XFS_IS_REALTIME_INODE(ip)) + xfs_update_stable_writes(ip); + /* * If there is no attribute fork no ACL can exist on this inode, * and it can't have any file capabilities attached to it either.
Update the per-folio stable writes flag dependening on which device an inode resides on. Signed-off-by: Christoph Hellwig <hch@lst.de> --- fs/xfs/xfs_inode.h | 8 ++++++++ fs/xfs/xfs_ioctl.c | 9 +++++++++ fs/xfs/xfs_iops.c | 7 +++++++ 3 files changed, 24 insertions(+)